RION-ANTIRION BRIDGE: FULL-SCALE TESTING OF SEISMIC DEVICES

Samuele Infanti1, Panayotis Papanikolas2, Giorgio Theodossopoulos2

1 FIP Industriale - ALGA - Freyssinet International Joint Venture (FIP R. & D. Dept. – Italy)2 Kinopraxia Gefyra - Greece

Keywords: Viscous Dampers, Fuse Restraints, Seismic Protection System, Energy Dissipation

1 ABSTRACT

The Rion-Antirion Bridge, which crosses the homonymous strait in Greece, is located in a regionprone to high intensity earthquakes and large tectonic movements generated by local active seismicfaults. The deck of the multi-span cable stay bridge is continuous and fully suspended (for its totallength of 2,252 meters) from four pylons (with spans 286m + 560m + 560m + 560m + 286m). Thebridge is designed to withstand seismic events with peak ground acceleration of 0.48g (2000 yearsreturn period) and tectonic movements between two piers of 2 meters in any direction. This waspossible through the use of an innovative energy dissipation system that connects the deck to the topof each pier and limits the movement of the deck during the specified earthquake, while dissipating theseismic energy.

The seismic protection system comprises fuse restraints and viscous dampers acting in parallel,connecting the deck to the piers. The restrainers are designed as a rigid link intended to withstand thehigh wind loads up to a pre-determined force corresponding to a value slightly higher than the windload ultimate limit state. Under the specified earthquake, the fuse restrainers are designed to fail,leaving the viscous dampers free to dissipate the energy induced by the earthquake into the structure.

The behavior and the requirements of the dissipation system were optimized with the help of anon-linear time history numerical analysis on a 3D model. It resulted on seismic devices (viscousdampers and fuse links) characterized by extreme design parameters in terms of force, velocity andstroke, with dimensions and weight never attempted before. The design assumptions and the actualbehavior of the seismic protection system had to be confirmed with extensive testing of full-scaleprototypes.

The aim of the present paper is to present the testing program and the results of the full-scaleprototype testing. The Viscous Damper Prototype tests were performed at the laboratory of theUniversity of San Diego California (USA), while the Fuse Restraints were tested at the FIP IndustrialeTesting Laboratory in Italy.

2 INTRODUCTION

The development of civil structures like bridges and buildings characterized by ever increaseddimensions has led to a confrontation of very complex problems linked to the imposition of dynamicactions by natural events. Windstorms and earthquakes can in fact compromise the function as wellas the safety of such structures.

In many cases, induced structural oscillations can be efficiently damped through the introduction ofdevices endowed with a constitutive force-velocity law (viscous dampers) that can furnish anothercapability to dissipate the energy associated with the aforesaid natural events.

In order to obtain the most efficient performance these devices shall be installed where themaximum relative movement between two adjacent structural elements is allowed.

The Rion-Antirion Bridge, located in an area prone to strong seismic events and windstorms, shallbe equipped with a damping system that for its characteristics and for the strict design requirementsentitles to put this structure in the next generation of seismic protected structures.

The seismic protection system comprises fuse restraints and viscous dampers acting in parallel,connecting the deck to the piers. The restrainers are designed to work as a rigid link in order towithstand the high wind loads. Under the specified earthquake, the fuse restrainers are designed to

fail, leaving the viscous dampers free to dissipate the energy induced by the earthquake into thestructure.

Four viscous dampers (Fmax 3500 kN, Stroke ±1750 mm) and one fuse restrainer (Fmax 10500kN) shall be installed at each pylons, while at the transition piers two viscous dampers and (Fmax3500 kN, Stroke ±2600 mm) and one Fuse Restrainer (Fmax 3400 kN) shall be considered.

These viscous dampers are of ever built dimensions and design capacity and shall be installed atthe pier to deck interfaces in order to reduce the transverse swing of the deck during a dynamic event.

They are designed to dissipate the energy introduced into the structure by the earthquake intoheat.

Requirements for a proper and safe seismic behavior are sometimes not in agreement with theeveryday service life of the structure. Thus, large structural displacements induced by moderateearthquakes or windstorms are avoided by an additional restraint system, that at the occurrence of amajor design event fails allowing the structure free to oscillate with its damping system.

This system is composed by Fuse Restraints installed in parallel to the dampers, so that the deck,when subjected to lateral loads not exceeding their design capacity, is linked rigidly to thesubstructure, while after failure it leaves the deck free to swing coupled to the dampers.

Figure 1 gives the general arrangement of the four viscous dampers and the restraint device atone pylon. The dissipation system connects in the transverse direction the fully suspended deck to thepylon base.

3 NON-LINEAR VISCOUS DAMPERS

FIP fluid viscous dampers are piston/cylinder devices that utilize fluid flow through orifices toabsorb energy. Orifices are situated in the piston head, which allow the fluid to move back and forthbetween the two chambers. The force generated by these devices is the result of a pressuredifferential across the piston head. These devices are equipped with two spherical hinges at bothends that keep the transmitted load aligned along their main axis. This detail is of major importance toprovide a reliable performance: it prevents the piston rod from bending and thus the sealing systemfrom failing. High strength steel components have been used for the vessel and the plated piston rodso as to withstand the actions imposed by a dynamic load. The anchoring details depend only on thestructure to which they are anchored: for example, the tang plate/clevis system illustrated in Figure 2.

In the last few years, FIP dampers have undergone major testing programs at FIP laboratory aswell as at independent facilities [3] [11] (Earthquake Engineering Research Center, Berkeley CA –USA, Boeing Testing Facility, Canoga Park CA – USA and Caltrans SRMD Testing Laboratory at the

Figure 1 Rion Antirion Bridge: Arrangement of Viscous Dampers

University of California – San Diego - USA), which has entitled the company to be a pre-qualifieddamper manufacturer for the Golden Gate Bridge Retrofit project as well as to enter the CALTRANS(California Department of Transportation) list of pre-qualified damper manufacturers. It should benoted that remarkable reaction stability has been achieved by dynamically cycling within a very wide

temperature range (-40°C ÷ +50°C) that guarantees proper behavior under any type of environmentalconditions.

A very important issue related to the utilization of the technology entails the correct numericalmodeling of the devices as integrated into the structural model and the performance verification of thefinished product. The latter topic will be amply discussed as follows.

The behavior and the requirements of the dissipation system were optimized with the help of anon-linear time history numerical analysis on a 3D model of the entire bridge. A parametric study onthe effect of the characteristics of the dissipation system demonstrated that the structural integrity

does not depend on the exact value of the non-linear exponent (α). See Table 1 for the stroke and theforce on the dampers for different cases of non-linear exponent. Despite this, strict specifications havebeen imposed with regards to the behaviour of the dampers in order to ensure a stable performance ofall devices.

Table 1 Parametric study – Effect of the non-linear exponent ‘α’

Non-linear exponent

‘α’

Maximum DynamicStroke (m)

Force(MN)

0.1 1.24 3.20

0.2 1.18 3.25

0.3 1.11 3.30

0.4 1.06 3.40

3.1 Modelling

The most appropriate mathematical model to represent the behavior of an FIP viscous damper isto use a Maxwell constitutive law characterized by a linear spring in series to a non-linear dash-potelement (see Figure 3). The first element represents the elasticity of the system that is mainly due tothe compressibility of the silicon fluid, while the second depicts its damping properties. The exponentthat characterizes the non-linearity of the response as a function of the velocity, is particularly

Rion-Antirion Main BridgeRion-Antirion Main Bridge

Viscous Damper Viscous Damper

PistonPiston

VesselVessel

HingeHinge

HingeHinge

Figure 2 Damper Main Components

important. Positive effects are undoubtedly related to low exponent dampers: e.g. in bridgeapplications, reduced pier deformation, lower expansion joint costs and cost effective bearings or inbuildings, maximum reduction of relative displacement at the isolation interface or in the braces.

FIP viscous dampers are designed to guarantee a 0.15 exponent so that, within a wide velocityrange, reaction is expected to be almost constant. This performance characteristic permits thedevices to initiate their damping reaction at low velocities, providing the maximum reduction of thesuperstructure displacement.

Total system elasticity, mathematically represented by the stiffness K, depends on thecompressibility of the fluid and it is particularly evident at the point of motion reversal, or at any rate, atthe point of transitory phases: commonly, it is a secondary effect and often it is not even computed.

Test 1 - Performance Benchmark (TA #3)

-800

-600

-400

-200

0

200

400

600

800

-250 -200 -150 -100 -50 0 50 100 150 200 250 300

Displacement (mm)

Fo

rce (

kN

)

Maxwell Model

K C, α

x(t)

150

0

.

)(

=

−=

=

α

ωα

K

FxCF

tsinxx

˙˙

Figure 3 Damper constitutive law equation and hysteresis loop

To evaluate the response of such a system, the differential non-linear equation representing itsconstitutive law needs to be integrated.

Looking at the plot in Figure 3, it can be noted that this typical hysteresis loop looks almostrectangular even if the displacement time–history is sinusoidal (thus velocity dependent). This is dueto the damper characteristic exponent (0.15). In fact, the behavior is almost independent from velocity.Thus total damper behavior can be similar to that of a spring in series to a ”perfectly plastic” elementcharacterized by a constant threshold force.

3.2 Viscous Dampers Full-Scale Testing

The following paragraphs present the main result of the testing activities performed at both FIPIndustriale Testing Laboratory and at the Seismic Response Modification Device (SRMD) TestingLaboratory of Caltrans at the University of California at San Diego [11] on a full-scale prototype of theviscous dampers to be utilized on the above mentioned bridge.

The prototype, characterized by a 3500 kN reaction at the maximum design velocity of 1.6 m/s andby a ±900mm stroke, was submitted by FIP Industriale on February 2002 for evaluation.

This viscous damper is deemed to be the largest ever-built device of its type (6.4 m pin to pin) andwas tested up to its maximum design conditions. The prototype is equal in every detail to the dampersdesigned for final installation with the exception of the stroke, which is shorter so as to fit into theexisting test rig.

3.2.1 Full-scale testing at FIP Industriale Laboratory – Italy

The above mentioned prototype underwent to two series of tests. The first one was performed atFIP Industriale Testing Laboratory, where the unit was tested for final tuning of its dampingcharacteristics before shipment to the UCSD laboratory for official testing in presence of the client(Kinopraxia Gefyra - Greece) and of the bridge design checker (Buckland & Taylor - Canada).

FIP Industriale Testing Laboratory is equipped with a power system providing for 630 kW at1200l/min thus resulting to be the most powerful in Europe for full-scale dynamic testing of seismicdevices.

Being the above unit of exceptional characteristics, the prototype was tested up to the maximumvelocity provided by the available system which resulted to be 0.2 m/s. In the following figure, the testresults are reassumed and it is evident how the damper behavior follows the theoretical constitutivelaw with a minimal deviation.

Figure

Figure 4 Prototype during testing at FIP Laboratory

Figure 5 Damper Constitutive Law

In Figure 4, the prototype is shown during dynamic testing at FIP testing Laboratory. For testingpurposes, the unit was pinned at one end to the test rig (yellow colored) while its piston rod wasconnected directly to the actuator through a 3000kN capacity load cell. The damper reaction wasmeasured through the mentioned load cell as well as indirectly from the pressure detected on theactuator through a pressure gauge. The displacement was measured through a magneto-strictivedisplacement transducer (see Figure 5 for testing results).

ActuatorTest Rig Frame

Servo-valveDamper

3.2.2 Full-scale testing at SRMD Testing facility – California, USA

The damper OTP 350/1800 was tested at the Seismic Response Modification Device (SRMD)testing facility of the University of California San Diego. The test matrix is reported in Table 2.

The specimen was horizontally installed in the testing rig, connected at one end to a reaction walland at the other end to the movable platen. The six degrees of freedom table is capable of uniquelevel of displacement, forces and velocities. For this specific application the platen was moved in thelongitudinal direction, with a displacement control loop able to maintain the components of motion inthe other directions at negligible level. Figure 6 shows the damper installed on the testing frame.

Figure 4 Prototype testing at SRMD Facility (UCSD)

A summary of results is presented in Table 3 for Thermal, Stroke Verification and Velocity Variationtests. Peak forces at different velocity levels are reported in Figure 7.

Table 2 Test Protocol Prototype FIP OTP 350/1800

Test#

TestName

Input Numberof cycles

Stroke Testing conditions(V = Peak Velocity)

1 Thermal Linear 1 ± 895 mm V<0.05 mm/s for 5 minutesIncrease velocity to 1mm/s up tocompletion of the displacement.

2 VelocityVariation

Sinusoidal 55532

± 300 mm

V=0.13 m/sV=0.40 m/sV=0.80 m/sV=1.20 m/sV=1.60 m/s

3 Full stroke&Velocity

Sinusoidalor step loading

1 ± 850 mm Vmax=1.6 m/s

4 Wear Linear 20000 ± 5 mm V=15 mm/sEvery hour change position of thepiston of about 100 mm

5 Velocityvariation

Sinusoidal 2 ± 300 mm Vmax=1.6 m/s

Maximum forces appear to be very symmetric in the all range of velocity. A difference of 7% wasrecorded at maximum speed (1.6 m/s) only for the first cycle. The second cycle of the same testshows instead a deviation of 1.6%. The comparison among peak forces of different cycles shows areduction of the peak force of 3.8% between fifth cycle and first cycle for test Velocity A (0.13 m/s). Forthe high speed tests the maximum force reduction is equal to 10.4%.

Table 3 Summary of Test Results

Test nameForce

1st LoopForce

2nd LoopForce

3rd LoopForce

4th LoopForce

5th Loop(kN) (kN) (kN) (kN) (kN)

Thermal (max.) 256.5 - - - -(min.) -111.4 - - - -

Stroke Verification 1 (max.) 1500.3 - - - -(min.) -1225.6 - - - -

Stroke Verification 2 (max.) 1003.9 - - - -(min.) -1019.2 - - - -

Velocity Variation A (max.) 2347.8 2315.2 2301.1 2292.2 2287.4(min.) -2273.7 -2242.8 2212.8 -2192.5 -2186.3

Velocity Variation B (max.) 2717.4 2730.4 2717.5 2702.3 2697.9(min.) -2674.7 -2671.9 -2657.1 -2647.3 -2648.0

Velocity Variation C (max.) 3004.4 3021.2 3016.4 3008.1 3009.0(min.) -2953.2 -2964.0 -2943.2 -2942.4 -2946.3

Velocity Variation D (max.) 3225.3 3215.9 3222.5 - -(min.) -3156.1 -3173.7 -3133.4 - -

Velocity Variation E (max.) 3722.9 3337.1 - - -(min.) -3461.5 -3281.6 - - -

Full Stroke & Velocity (max.) 3704.9 - - - -(min.) -3425.0 - - - -

Velocity Variation F (max.) 3590.5 3310.4 - - -(min.) -3701.5 -3371.6 - - -

The calculated energy dissipated per cycle (EDC) indicates a reduction of 5.2% between cycles forthe high speed test (Velocity Variation E).

Wear tests were completed with 10 sets of 2250 cycles, at constant velocity of 0.015 m/s and10 mm total stroke.

Figure 5 Damper Constitutive Law

The force displacement response of the damper is reported in Figure 8 for the test at 1.2 m/s peakvelocity.

Figure 6 Damper Hysteresis Loop

Thermocouples were installed both internally and externally the damper body in order to monitortemperature rise during and after the motions. Air and nitrogen gas was used, in a cooling box, torestore the ambient temperature on the damper before a new test.

Temperature rise was recorded for each test. The maximum increase took place at the end of thetest Velocity Variation B, with 40 degrees Celsius recorded from the sensor installed internally to thedamper.

All the test results were deemed to be in agreement with the design specifications.It is worth to stress how the test results obtained at the SRMD facility agreed very well with the

measurements performed at FIP testing laboratory and how the damper behavior resulted to providefor stable reaction in a very wide velocity range (0.002 – 1.6 m/s).

The extrapolation of the damper reaction in the range of velocity 0.13-1.6 m/s from the testperformed at FIP laboratory up to 0.2 m/s differed from the measured reaction at the SRMD facility byonly few percent point. This result demonstrates how much predictable is the damper behavior withinthe full test range.

4 FUSE RESTRAINTS

Fuse restraints shall provide for a stiff link between the deck and the substructure for the lateralservice load as well as for low occurrence high-intensity windstorms. As anticipated, only at theoccurrence of high intensity events they are designed fail allowing the damping system to workproperly.

They are located at the main piers as well as at the transition piers. The main pier units aredesigned as single units equipped at their ends with spherical hinges. This configuration allows for thedesign rotations as well as for a correct alignment of the load along the device axis for any deckposition. The element failing at the reaching of the desired design load – the so called Fuse Element -is installed in the middle of the unit, for a general configuration of the unit see Figure 9.

The units to be installed on the main piers are characterized by a failure load of 10500kN. Similarlythe units installed at the transition piers, that are installed as components of the dampers, arecharacterized by a failure load of 3400kN. The design tolerance on the failure load of the unitsrequires a precision of ±10%: a very strict design performance for such a high capacity units.

Furthermore, a main design complication came from the need of minimizing the internal forces ofthe laterally restraint deck induced by tectonic movements originated from seismic fault located underthe bridge. Thus, the units located on the main piers are equipped with a system designed to allow forlength adjustment. This operation is performed when a certain load level is constantly applied to thelink, so a load cell monitoring the load on the unit is required. Other requirements were that the unitsshall not disassemble after failure of the main component and that it has to be equipped with a loadcell in order to monitor the load acting on the unit. As part of the monitoring system of the bridge, thiswill allow to identify the moment at which re-adjustment of the deck is required.

Being their first function to withstand the every day actions (service loads), a test has beenrequired to evaluate the fatigue life as well as any influence on the failure strength.

Figure 7 Fuse Restraint General Configuration

4.1 Testing program

The testing program was carried out on two full-scale prototypes of fuse element for eachtypology.

It consisted on a first test performed on one unit increasing monotonically the load up to failure,while a second test was performed on the other prototype imposing two millions of cycles at a loadlevel equal to 10% of the design failure load and then increasing monotonically the load up to failure.

Failure test and fatigue test were carried out on different test rigs, the first one is a 8000 kNcapacity test rig commonly used for bearing test while the second one is a 3000kN dynamic test rig:the same used for damper testing.

Testing results have shown that both prototypes failed within the design tolerance (±10% of thetheoretical load). The very slight difference, about 1%, between the two measured failure loads canjust confirm that the design fatigue load cannot be considered as effecting the ultimate capacity of thefuse restraints. In the following table, all the testing results are presented. Figure 10 shows typicalgraphs obtained for both the 3400kN and 10500kN units.

Table 4 Test Results

Device Type Failure Load Capacity(kN)

Tolerance Range(kN)

Measured Load(kN)

Deviation (%)

SR340 3400 3060-3740 3545 +4.3

SR340 3400 3060-3740 3591 +5.6

SR1050 10500 9450-11550 11000 +4.8

SR1050 10500 9450-11550 NA NA

Figure 11 shows the SR1050 fuse element testing configuration during the fatigue test.

SR340 - Failure Test SR1050 - Failure Test

Figure 8 Failure Test

Figure 9 SR1050 Fuse Element During Fatigue Test

5 CONCLUSIONS

Testing of full-scale hydraulic damper and fuse elements were performed in order to confirm thecharacteristics and the main design assumptions of the seismic dissipation system used for theRion-Antirion Cable stayed bridge.

Tests performed on Fuse elements verified positively the design assumption showing that it is

possible to achieve a very tight tolerance (±10%) on the predicted ultimate capacity even when theunits are designed for a very high failure load (10500kN).

The Viscous Damper prototype demonstrated very stable behavior even when tested underdynamic conditions, requiring power dissipation higher than design parameters.

The measured energy dissipation and viscous reaction were always well within the designtolerance of ±15% of the theoretical design parameters.

Testing aimed to represent the effect of structural vibrations induced by traffic and/or of movementinduced by structure thermal expansion did not produce any appreciable change in the damperbehavior.

Full-scale testing of large size dissipating devices proved to be a real challenge compared toprevious testing experiences as it pushed the limits of equipment available worldwide.

Experimental results on full-scale prototypes give a high degree of confidence on seismic devicesas a means to protect large and important structures such as the Rion-Antirion Cable Stay bridge.

ACKNOWLEDGEMENT

The authors wish to express their sincere gratitude for the active and fruitful cooperation renderedby Mr. Gianmario Benzoni of the UCSD Laboratory, as well as the indefatigable support furnished byMr. Renato Chiarotto of FIP Technical Department, Mr. Pierluigi Galeazzo Testing LaboratoryManager of FIP Industriale and his staff.

A remarkable thank is addressed to Mr. Agostino Marioni (ALGA) and to Mr. Philippe Salmon(Freyssinet) for their cooperation within the JV operations.

REFERENCES

[1] Soong T.T., Dargush G.F., ”Passive Energy Dissipation Systems in Structural Engineering”,Wiley, Chichester – England, 1997, 356 pp.

[2] Technical Report “Guidelines for Testing of Seismic Isolation and Energy Dissipating Devices”, Highway Innovative Technology Evaluation Center, 1998.

[3] Technical Report “Evaluation Findings of FIP Industriale Viscous Dampers according to theHITEC testing protocol”, Rocketdyne Division of Boeing, Canoga Park – CA USA, 2000.

[4] Infanti S., “Seismic Protection of Seohae Bridge Approaches trough Viscous Dampers”, Cable-Supported Bridges – Challenging Technical Limits, IABSE Conference, Seoul - Korea, 2001.

[5] Technical Report “Evaluation Findings for FIP – Energy Absorption Systems, LLC SliderBearings ”, Highway Innovative Technology Evaluation Center, 1998.

[6] Virtuoso F., Guerrero L., Azevedo J., “Modelling the Seismic Behavior of Bridges With ViscousDampers”, 12th World Conference of Earthquake Engineering, Auckland – New Zealand, 2000.

[7] Infanti S., Colato G.P., Castellano M.G., “Structural Devices for Bridge Engineering: theChallenge Goes On”, Bridge Engineering Conference, Sharm El-Sheikh – Egypt, 2000.

[8] Skinner R.I., Robinson W.H., McVerry G.H., ”An Introduction to Seismic Isolation”, Wiley,Chichester – England, 1993, 354 pp.

[9] Infanti S., Castellano M.G., “ Viscous Dampers: a Testing Investigation according to the HITECProtocol”, Fifth World Congress on Joints, Bearings and Seismic Systems for ConcreteStructures, Rome – Italy, 2001.

[10] Teyssandier J.P, Combault J, Morand P., “The Rion-Antirion Bridge Design and Construction”12th World Conference of Earthquake Engineering, Auckland – New Zealand, 2000.

[11] Benzoni G., Seible F, “Rion-Antirion Bridge - Prototype Tests of Damper FIP OTP 350/1800”Caltrans SRMD Test Facility Report N. 2002/01, San Diego CA - USA, 2002.

[12] Infanti S., Castellano M.G., Benzoni G. “Non-Linear Viscous Dampers: Testing and RecentApplications”, ATC 17-2A Seminar on Response Modification Technologies for Performance-Based Seismic Design, Los Angeles CA - USA, 2002.